Next: About this document ...
Up: SEGMENTATION
Previous: Correctness
- Problem: word segmentation
- Step 0 (formalize): unigram model, find maximum probability segmentation
- Step 1 (devise algorithm): table-based approach (DP)
- Step 2 (correctness): follows from equation for probability
- Step 3 (time): time per cell times number of cells (O(n2))
- refine: better formal model, better data structure, worse formal model!
Guangwei Yuan
12/4/1999